改进的共享型最近邻居聚类算法

聚类效果往往依赖于密度和相似度的定义，并且当数据的维增加时，其复杂度也随之增加。该文基于共享型最近邻居聚类算法SNN，提出了一种改进的共享型最近邻居聚类算法RSNN，并将RSNN应用于高速公路交通数据集上，解决了SNN算法在“去噪”、孤立点和代表点的判断、聚类效果等方面的不足之处。实验结果表明，RSNN算法比SNN算法在时空数据集上具有更好的聚类效果。

Clustering results often depend on density and similarity critically, and its complexity often changes along with the augment of sample dimensionality. This paper refers to classical shared nearest neighbor clustering algorithm (SNN) and refined shared nearest neighbor clustering algorithm (RSNN). By applying this RSNN algorithm on freeway traffic data set, we settled several problems existed in SNN algorithm, such as outliers, statistic, core points, computation complexity and so on. Experiment results prove that this refined algorithm has better clustering results on multi-dimensional data set than SNN algorithm.